I have mine only internal so i haven’t ran into that. But check console. You mention mobile so if you’re on android you can hook it up to your pc and use debugging through chrome.. In the past I’ve had success looking at error messages to see why my requests were failing. Usually because i wasn’t passing headers correctly.

I use symfonium and it looks like it let’s you pass custom headers if needed. Good luck

Less about customizations and more just it doing what i want, and not doing things i don’t want. When you build it all from the ground up then you don’t have surpise bloat or walls to work around/within.

But most of my customizing from what people use probably would be around my dev environments. Things like rebuilding python libraries to support my gpu are fairly trivial in arch when i need to deviate from releases available through package managers (aur/pypi). Another thing was setting up my data science environments to share some core libraries but venv the rest.

It’s a hard question to answer though because fundamentally I’m just using the computer how i want to use it. When you say customization it sounds like you are expecting me to do things differently than other people and really it’s just like i said earlier-- doing things i want it to do, and not doing things i don’t want it to do. And I’m not really sure what walls other people are stuck behind for me to know what I’m doing differently. I just find a problem, fix it, and move on

For me it mostly just came down to years of frustration combating windows to do what i wanted. Arch offers the level of control for me to set things up the way i like them. Was it harder to set up initially than another distribution? Yeah. But it was a worthwhile trade off

Neat project! And welcome to lemmy!

(Not mint)* On arch i used the arch install script, selected the nvidia drivers, and it just worked. I did have to spend some time making sure sure my nvidia gpu was my primary gpu and not my integrated graphics (cpu), but that was the biggest hurdle

That #1 is crucial. I see a lot of people get stuck in tutorial hell or burn out from doing other people’s projects. Some tutorials are okay if you’re just starting out but at some point switching to your own projects and challenging yourself is necessary

And since OP mentioned being on/off, i would also just say be consistent. Dedicate some time to work on your own projects so you’re not forgetting stuff before it really sticks

I switched January this year.

1. Windows 10 end of life was on the horizon
2. Programing on windows was a lot of hoops to jump through and i had heard Linux would be better
3. Didn’t want windows 11/copilot.

I struggled with that but for me i treated it as one I’ve been most hyped about this past year

I stream music via my own server. I use navidrome. I use their Web UI for PC and then symfonium for mobile.

But yeah you’re right, it’s cheaper for me to buy a few CDs every now and again than pay a subscription

I selfhost navidrome for the music streaming (+symfonium app for mobile). Multi user and multi library support.

For music tagging itself ive used beets, picard, and kid3 (kde). Currently I am liking picard the most. It took a little bit of learning but less than beets

Does someone have the tldr?

/s

Expired domains first can be bought by registrars and then they might sell or auction it off. For instance godaddy will scoop up a lot of domains and auction them off even if it was registered somewhere else first. And unfortunately a surname.tld probably will invite domain squatters to try to get it and then charge much more for it

You can look into something like dropcatch which they will try to get the domain for you before another registrar gets it. Look into their backorder service and just check the timing to make sure they still can try to get it.

Regularly check the whois info (via icann lookup) to see which registrar currently has it which can help you determine if it has gone to an auction.

Who needs access to these private repos? There’s always raw git (has a web server if needed). That’s what I’ve been doing since moving to codeberg for my public projects and eventually i might set up a private forgejo server.

Sourcehut also offers public, private, and unlisted repos

There are a few options

• cloudflare pages if you’re not against cloudflare.
• sourcehut pages
• codeberg pages - though they list it at in maintenance mode atm so i wouldn’t put anything critical on there

I personally would just avoid netflify (something about them sending a bill for a DDoS that wasn’t mitigated gave me a bad taste).

Ofc nothing making you need to move right now, but id explore options so you have a gameplan

Giscus / utteranc.es use github discussions/ issues to do comments so that might be an option

For one of my projects i setup remark42 which does allow anonymous comments. You can also set it up to allow logging in to a few different platforms. 7 months and no issues.

Isso is another one i had looked at which can do comments without an acct

There are cuda packages for arch. I can confirm they work.

https://wiki.archlinux.org/title/GPGPU

Edit- to add i haven’t found arch’s limit, it feels like i am bound only by my own limit in my time, ability, and willingness to tinker with my setup. arch itself is widely supported, it has many official packages through pacman, and then additional through the “arch user repository” (AUR) so chances are most of the things you want or need have a package that can be installed with an AUR helper (like yay/paru which install from both pacman and AUR). In other distros you get more of a one-size fits all and you lose some of that ability to change things whereas arch you’re giving a minimal setup and left to build the system to your liking. It does take more time and expertise than other distros but it does give you more control. For me the trade off was an easy decision, but it’s not something i blanket recommend to everyone.

Ah, my apologies!

Codeberg has been stable enough for my small usage. It does have a CI, woodpecker, that requires manual approval. I haven’t used their CI yet

Tflops is a generic measurement, not actual utilization, and not specific to a given type of workload. Not all workloads saturate gpu utilization equally and ai models will depend on cuda/tensor. the gen/count of your cores will be better optimized for AI workloads and better able to utilize those tflops for your task. and yes, amd uses rocm which i didn’t feel i needed to specify since its a given (and years behind cuda capabilities). The point is that these things are not equal and there are major differences here alone.

I mentioned memory type since the cards you listed use different versions ( hbm vs gddr) so you can’t just compare the capacity alone and expect equal performance.

And again for your specific use case of this large MoE model you’d need to solve the gpu-to-gpu communication issue (ensuring both connections + sufficient speed without getting bottlenecked)

I think you’re going to need to do actual analysis of the specific set up youre proposing. Good luck

The table you’re referencing leaves out CUDA/ tensor cores (count+gen) which is a big part of the gpus, and also not factoring in type of memory. From the comments it looks like you want to use a large MoE model. You aren’t going to be able to just stack raw power and expect to be able to run this without major deterioration of performance if it runs at all.

Don’t forget your MoE model needs all-to-all communication for expert routing